pubs.acs.org/NanoLett
Stepwise Nanopore Evolution in One-Dimensional Nanostructures Jang Wook Choi,†,⊥ James McDonough,†,⊥ Sangmoo Jeong,‡ Jee Soo Yoo,† Candace K. Chan,§,| and Yi Cui*,† †
Department of Materials Science and Engineering, ‡ Department of Electrical Engineering, and § Department of Chemistry, Stanford University, Stanford, California 94305 ABSTRACT We report that established simple lithium (Li) ion battery cycles can be used to produce nanopores inside various useful one-dimensional (1D) nanostructures such as zinc oxide, silicon, and silver nanowires. Moreover, porosities of these 1D nanomaterials can be controlled in a stepwise manner by the number of Li-battery cycles. Subsequent pore characterization at the end of each cycle allows us to obtain detailed snapshots of the distinct pore evolution properties in each material due to their different atomic diffusion rates and types of chemical bonds. Also, this stepwise characterization led us to the first observation of pore size increases during cycling, which can be interpreted as a similar phenomenon to Ostwald ripening in analogous nanoparticle cases. Finally, we take advantage of the unique combination of nanoporosity and 1D materials and demonstrate nanoporous silicon nanowires (poSiNWs) as excellent supercapacitor (SC) electrodes in high power operations compared to existing devices with activated carbon. KEYWORDS Nanopore, nanowire, supercapacitor
P
orous materials have been utilized in numerous applications including catalysts,1,2 molecular sieves,3 chemical and biological sensors,4 low k dielectric layers in microelectronic devices,5 hydrogen storage materials,6 supercapacitor electrodes,7 contrast agents for magnetic resonance imaging (MRI),8 and zeolite-dye microlasers.9 While template-based syntheses10,11 have been major methods in producing porous materials with highly ordered pore structures and large surface areas (>1000 m2/g), the dealloying process has also been utilized to produce a variety of porous materials and expand the materials pool.12–14 The dealloying process provides unique, complementary advantages to the template-based process; the dealloying process is compatible with various nanomaterials, especially those grown directly on substrates. Therefore, various electrochemical devices whose efficiencies are directly correlated to the yields of surface reactions should benefit substantially from these porous nanostructures. Also, the procedure is simpler because it can be done at room temperature and requires only a couple of reaction steps. Despite these advantages, pore size control in the dealloying methods has been limited to bulk-scale metal alloys focusing on thermal annealing13 and alloying with additional elements.15 Hence, it must be meaningful to understand pore evolution in various materials beyond metals and thus develop the capability of controlling the porosities of those materials. In particular, nanopore generation and their size control in
nanomaterials should prove to be very useful in diverse applications. In this study, established Li-battery cycles are used to generate porous, one-dimensional (1D) nanostructures such as porous zinc oxide nanorods (ZnONRs), silicon nanowires (SiNWs), and silver nanowires (AgNWs). We chose these materials because their bonding characteristics are completely different such that the dealloying process is expected to produce nanopores in distinct ways. Because of the small dimensions of the nanowires and nanorods, the pore formation can be observed clearly using electron microscopes. Moreover, the pore evolution can be monitored quantitatively at its various stages of pore growth. Among many potential applications that can take advantage of these wellbound nanostructures with increased surface areas, supercapacitors are demonstrated here as a useful example. Detailed descriptions for nanostructure growth processes, battery sample preparation, and potential profiles for the battery cycles are presented in the Supporting Information (Figures S1 and S2). In the anode during a Li-battery cycle, Li is inserted during the charging process and is extracted during the discharging process. During the charging process, ZnO goes through the following conversion reaction followed by an alloying reaction
2Li + ZnO T Li2O + Zn
(1)
* To whom correspondence should be addressed. E-mail:
[email protected].
Zn + Li T LiZn
(2)
|
Present address: Department of Chemistry, University of California, Berkeley, CA 94720.
⊥
These authors contributed equally to this work. Received for review: 01/23/2010 Published on Web: 03/25/2010 © 2010 American Chemical Society
Once zinc oxide is lithiated in reaction 1, zinc nanograins are formed in a lithium oxide matrix.16 If the lithiation 1409
DOI: 10.1021/nl100258p | Nano Lett. 2010, 10, 1409–1413
for both materials. This observation is consistent with the significant pore size increases in the pore size distribution (PSD) data during the same Li-battery cycling period. The stepwise pore size increase can be interpreted as a surface energy reduction brought about by larger pore sizes. Similarly to Ostwald ripening19 in analogous nanoparticle cases20 in which nanoparticles are likely to agglomerate in order to minimize surface energy, pores are also likely to increase their widths or merge together to form bigger ones, resulting in a reduction of the surface energy. This is not only the first observation directly related to Ostwald ripening-like phenomena in pore evolution but also provides a clear indication that nanopores can be generated inside diverse 1D nanomaterials and their pore sizes are tunable by the number of electrochemical cycles. A similar method of producing porous metal oxides by using Li-battery cycles has been reported.21 However, their materials were limited to metal oxides, and adjustable electrochemical parameters to control the porosities were not explored. Second, pores evolve in different fashions depending on the structural properties of lithiated active materials. While ZnONRs show dramatic increases in pore volumes and BET surface areas after just one Li-battery cycle, SiNWs show gradual increases throughout Li-battery cycling. In the case of ZnONRs, the most feasible mechanism is that a Kirkendall-like effect between the fast-diffusing Zn in the Zn nanograins and the relatively slowly diffusing O2- ions in the Li2O matrix leads to a nanograin-to-pore transition. Hollow structures in nanoparticles driven by a similar Kirkendall effect between fast-diffusing cobalt ions and slowly diffusing oxide or sulfide have been reported.22 Our study here is similar to that case. The zinc nanograins function as templates, directing the pore formation. This hypothesis is further supported by the similar size (1-8 nm) between the nanograins and the pore widths and the narrower PSD. On the other hand, in the case of SiNWs, unstable Si dangling bonds surrounding vacancy sites that could appear in the discharging process might preclude efficient pore formation during their initial cycles. However, the poor diffusivities of Si atoms lead to pore formation on the NW surfaces in the early cycles and eventually further penetration toward the core in the later cycles, as indicated by TEM (Figure 3B) and SEM (Figure 4A) images. Therefore, this template-free mechanism results in a gradual increase in pore volume and a broader PSD. We also tested silver nanowires (AgNWs) to see how atomic diffusion affects the pore evolution. Because of the soft nature of silver, pore formation in the nanostructures is not as efficient as in SiNWs and a pore size decrease during 5 to 10 Li-battery cycles was observed, a trend opposite to those of ZnONRs and SiNWs (see Supporting Information Figures S4 and S5 for detailed data and descriptions). Because each porosity data point shown in Figure 2 and Supporting Information S4 is produced from more than 20 Li-battery cells, all of the porosity data should be highly representative.
FIGURE 1. Schematic illustration of pore generation by the alloying/ dealloying process. As-grown single-crystalline zinc oxide nanorods or silicon nanowires expand upon electrochemical Li ion insertion. As Li ions are extracted from these 1D nanomaterials, the nanostructures shrink but to volumes larger than the original. The increased volumes suggest pore formation inside the nanostructures.
proceeds further, the zinc nanograins become lithiated (reaction 2). The conversion reaction (reaction 1) is confirmed by potential profiles (the plateau at ∼0.5 V, Supporting Information Figure S2) consistent with previously reported data17,18 as well as high resolution transmission electron microscopy (HRTEM) data (Supporting Information Figure S3). On the other hand, Li-battery cycles in silicon are based solely on an alloy-dealloy reaction
Si + 4.4Li T Li4.4Si
In the charging process (lithium insertion), the active electrode materials, ZnONRs and SiNWs in our case, undergo volume expansion. Theoretically, ZnO can take 3.0 Li per ZnO, and the volume upon full lithiation is about 163% of the original.17 Si can take up about 4.4 Li per Si, and its volume upon the full lithiation is even larger, up to 400% of the original. As Li is extracted in the discharging process, the expanded active materials shrink, but the final volume at the end of the discharging is still larger than the original, which indicates formation of void space inside the active materials (Figure 1). Thus, the pore formation depends directly on the volumetric expansion of the active materials in the charging process and the atomic diffusion of the active materials in the discharging process. When the atomic diffusion of active materials is slow during the discharging process, the active materials tend to form pores more efficiently because the slow diffusion prohibits the active materials from returning to their original structures. Instead, thermodynamically metastable structures with pores generated inside are preferred. Several interesting features in the pore evolution are noteworthy. First, we demonstrate that pore sizes are tunable by the number of Li-battery cycles. As shown in Figure 2, after every Li-battery cycle, pore sizes increase in a stepwise manner for both ZnONRs and SiNWs. The pore size increases are indicated by faster increases of pore volumes compared to those of BET (Brunauer-Emmett-Teller) surface areas (Figure 2B) as well as pore size distributions (Figure 2C). In fact, as the number of Li-battery cycles increase from 5 to 10 cycles, the BET surface areas decreases © 2010 American Chemical Society
1410
DOI: 10.1021/nl100258p | Nano Lett. 2010, 10, 1409-–1413
FIGURE 2. Pore size increases with Li-battery cycles. (A) The nitrogen adsorption and desorption isotherms measured at 77 K as a function of the number of Li battery cycles. Increased pore volumes at larger numbers of Li battery cycles are observed for both ZnONRs and SiNWs. (B) The BET surface area and pore volume changes for different numbers of Li-battery cycles. The increase in pore volume is more abrupt than the increase in BET surface area for both materials. Very small pore volumes at 0 Li-battery cycles indicate that interwire spacing is negligible in our measurement range (
